Petr Hála

> This shouldn't lose space. The only thing different that other similar tests is that I precreated the keyspace with initial tablets = 128. Could this be explained by major...

> Probably, but other reclaim tests are exactly the same, except for that part and it doesn't happen there > I will try again without it. I do not think...

@roydahan @fruch Let me know if you have questions about the general idea and if you think this is a good direction to take

> one more drawback of it, that it can't be disabled for release branches. i.e. the original idea is that we can work by the nemesis list, and by the...

> There is always a possibility of ignoring it, you are right that it is harder to do with precommit than with GH actions, but given that we never implemented...

> again the problem isn't about one extra commit, but one us having a ability to pin down the list of nemesis used in a release, which is the whole...

> @pehala > > one of the commits fails cause `s/improveme/improvement/` > > it can be added to the jscommitlint configuration, if you want > Nah this is error on...

> Generating jobs from these files is also manual process and requires running `hydra create-nemesis-pipelines --backend docker --backend aws --backend azure --backend gce`, which requires AWS login, even though it...

v2: Rework the PR * Drop the new tool, integrate everything into sct.py * Simplify changes * Remove precommit changes I also updated the cover letter. I would still like...

The checks are run from master (due to `pull_request_target`) and thus the change to the workflow file is not propagated.